1. Field of the Invention
The present invention relates to a semiconductor device and a driving method thereof. Specifically, the present invention relates to an MOS sensor device that has an image sensor function and to a driving method of the same.
2. Description of the Related Art
In recent years, the use of information equipment such as a personal computer has spread widely, and hence the demand to read (store) various information in the personal computer, etc. as electronic information is rising. Therefore, replacing the conventional silver salt camera, a digital still camera or a scanner, which is used as a means of reading information printed on paper, are in the spotlight.
An area sensor in which the pixels are arranged in a two-dimensional way is used in the digital still camera. In the scanner, a copier machine, etc., a line sensor in which the pixels are arranged in a one-dimensional way is used. In the case of using the line sensor to read a two-dimensional image, signals are read while moving the line sensor.
A CCD type sensor is mainly used as the image sensor in these types of image reading equipments. In the CCD type sensor, a photoelectric conversion is carried out in a photo diode of each of the pixels and then the CCD is used to read those signals. However, an MOS type sensor that is formed by using a single crystal silicon substrate is showing signs of popularization in a part of the technical field thereof by using factors such as the incorporation of a peripheral circuit, manufacturing it into one chip, its suitability for a real time signal process, and its low consumption power as weapons. Further, the manufacture of an MOS type sensor by using a TFT that is formed on a glass substrate is being developed at the research level. In the MOS type sensor, the photoelectric conversion is carried out in the photo diode of each of the pixels, whereby the signals of the respective pixels are read out by using a switch that is formed by an MOS transistor.
As a pixel structure of the MOS type sensor, various types are being developed. The various types of pixel structure of the MOS type sensor can be largely categorized into two types, that is, a passive sensor type and an active sensor type. The passive sensor is a sensor in which a signal amplitude element is not incorporated into the respective pixels whereas the active sensor is a sensor in which a signal amplitude element is incorporated into the respective pixels. The active sensor has an advantage over the passive sensor in that it is strong against noise because the signals are amplified in each of the pixels.
Shown in FIG. 2 is an example of a circuit of a pixel in the passive sensor. A pixel 10005 is composed of a switching transistor 10001 and a photo diode 10004. The photo diode is connected to a power source standard line 10006 and to a source terminal of the switching transistor 10001. Agate signal line 10002 is connected to a gate terminal of the switching transistor 10001, and a signal output line 10003 is connected to a drain terminal thereof. Photoelectric conversion takes place in the photo diode 10004. In other words, an electric charge is generated in response to the incidence of light, whereby the electric charges are accumulated therein. Then the switching transistor 10001 is made into conductive by controlling the gate signal line 10003 to thereby read out the electric charge of the photo diode 10004 through the signal output line 10003.
There are various kinds of pixel structure of the active sensor. Pixel structures such as a photo diode type and a photo gate type and their operations are introduced in IEDM95: p. 17: CMOS Image Sensors, Electric Camera On a Chip or in IEDM97: p. 201: CMOS Image Sensors—Recent Advances and Device Scaling Considerations. In the ISSCC97: p. 180: A ¼ Inch 330 k Square Pixel Progressive Scan CMOS Active Pixel Image Sensor, the pixel structure is categorized from the perspective of a selecting method of the pixel. That is, a case of whether to use a transistor or a capacitance as a selecting element is described therein. Thus, there are various types of structures regarding the number of transistors for forming one pixel. A general description of the CMOS type sensor is broadly introduced in the JIEC Seminar: Development Prospects of the CMOS Camera: Feb. 20, 1998. In the description thereof, a logarithm conversion form, which outputs a signal of the logarithm of light density by connecting a gate electrode and a drain electrode of a resetting transistor, is also explained.
As shown in FIG. 3, a pixel structure of the active sensor that is mostly adopted is a type that is composed of three N channel transistors and one photo diode, thereby forming one pixel 308. A P channel side terminal of a photo diode 304 is connected to a power source standard line 312, and an N channel side terminal of the photo diode 304 is connected to a gate terminal of an amplifying transistor 306. A drain terminal and a source terminal of the amplifying transistor 306 are connected to a power source line 309 and to a drain terminal of a switching transistor 301, respectively. A gate terminal of the switching transistor 301 is connected to a gate signal line 302 while a source terminal thereof is connected to a signal output line 303. Agate terminal of a resetting transistor 307 is connected to a reset signal line 306. A source terminal and a drain terminal of the resetting transistor 307 are connected to the power source line 309 and a gate terminal of the amplifying transistor 306, respectively.
In the case of an area sensor, not only one pixel 308 is connected to one signal output line 303, but also a plurality of pixels are connected thereto. However, one biasing transistor 311 is connected per signal output line 303. Agate terminal of the biasing transistor 311 is connected to a bias signal line 310. A source terminal and a drain terminal of the biasing transistor 311 are connected to the signal output line 303 and to a biasing power source line 313.
Next, a basic operation of the pixel 308 will be explained.
The resetting transistor 307 is first made into a conductive state. Because the P channel side terminal of the photo diode 304 is connected to the power source standard line 312, whereby the photo diode 304 becomes a state in which the N channel side terminal is electrically connected to the power source line 309, an inverted bias voltage is applied to the photo diode 304. Hereinafter, the operation of charging the N channel side terminal of the photo diode 304 until its electric potential is equivalent to the electric potential of the power source line 309 will be referred as “reset”. Thereafter, the resetting transistor 307 is made into a non-conductive state. When light is being irradiated to the photo diode 304, an electric charge is generated due to a photoelectric conversion. Therefore, as time elapses, the electric potential of the N channel side terminal of the photo diode 304, which has been charged up to the electric potential of the power source line 309, gradually becomes smaller because of an electric charge that was generated by the light. Then after a fixed period of time has passed, the switching transistor 301 is made into a conductive state, whereby a signal is output to the signal output line 303 through the amplifying transistor 306.
However, at the time the signal is being output, an electric potential is applied to the bias signal line 310 to cause a current to flow in the biasing transistor 311. Therefore, the amplifying transistor 306 and the biasing transistor 311 operate as the so-called source follower circuits.
An example of the most basic source follower circuit is shown in FIG. 4. In FIG. 4, the case of using the N channel transistor is described. Although a P channel transistor can be used to construct the source follower circuit, a case of using an N channel transistor is shown in FIG. 4. A power source electric potential Vdd is applied to an amplifier side power source line 403. A standard electric potential 0V is applied to a bias side power source line 404. A drain terminal of an amplifying transistor 401 is connected to the amplifier side power source line 403 while a source terminal thereof is connected to a drain terminal of a biasing transistor 402. A source terminal of the biasing transistor 402 is connected to the bias side power source line 404. A bias electric potential Vb is applied to a gate terminal of the biasing transistor 402. Therefore, a bias current Ib flows in the biasing transistor 402. The biasing transistor 402 basically operates as a fixed electric current source. A gate terminal of the amplifying transistor 401 serves as an input terminal 406. An input electric potential Vin is thus applied to the gate terminal of the amplifying transistor 401. A source terminal of the amplifying transistor 401 serves as an output terminal 407, and therefore an output electric potential Vout is applied to the source terminal of the amplifying transistor 401. At this point, the relationship of the input/output of the source follower circuit becomes Vout=Vin−Vb.
In the case of comparing the circuit configurations of FIG. 3 and FIG. 4, the amplifying transistor 306 corresponds to the amplifying transistor 401, and the biasing transistor 311 corresponds to the biasing transistor 402. Because it is assumed that the switching transistor 301 is in conductive, it can be observed that a switching transistor is omitted in FIG. 4. The electric potential of the N channel side terminal of the photo diode 304 corresponds to the input electric potential Vin (the gate electric potential of the amplifying transistor 401, that is, the electric potential of the input terminal 406). The electric potential of the signal output line 303 corresponds to the output electric potential Vout (the source electric potential of the amplifying transistor 401, that is, the electric potential of the output terminal 407).
Therefore, in FIG. 3, if the electric potential of the N channel side terminal of the photo diode 304 is Vpd, the electric potential of the bias signal line 310, that is, the bias electric potential is Vb, the electric potential of the signal output line 303 is Vout, and the electric potential of the power source standard line 312 and the bias side power source line 313 is 0V, then the relationship becomes Vout=Vpd−Vb. Accordingly, when the electric potential Vpd of the N channel side terminal of the photo diode 304 changes, then Vout also changes. As a result, the change of the Vpd can be output as a signal and the light intensity can thus be read.
The basic operation of the source follower circuit is one as described above. However, the operating principle of the source follower circuit will be explained next in detail because it is needed for explaining the operation of the present invention. For simplification, it is assumed that the sizes and characteristics of the amplifying transistor and the biasing transistor are the same in the explanation here. Further, an electric current characteristic of the transistors is an ideal one, that is, even if a voltage between the source and the drain changes, it is assumed that an electric current value in a saturated region does not change.
First, as shown in FIG. 4, the bias electric potential Vb is applied to the gate terminal of the biasing transistor 402. In the case the biasing transistor 402 operates in the saturated region, the electric current Ib flows therein as shown in FIG. 5. On the other hand, the same amount of electric current will flow in amplifying transistor 401 and the biasing transistor 402 under a fixed normal state because both transistors are connected in series. Therefore, when the electric current Ib is flowing in the biasing transistor 402, the electric current Ib is also flowing in the amplifying transistor 401. In order to cause the electric current Ib to flow in the amplifying transistor 401, it is necessary to make the voltage Vgs between the gate and the source of the amplifying transistor 401 equivalent to the bias electric potential Vb.
Thus, the output electric potential Vout in the source follower circuit is obtained. The amount of electric potential of the output electric potential Vout that is lower than the input electric potential Vin is equal to only that of the voltage Vgs between the gate and the source of the amplifying transistor 401. Therefore, the input/output relationship becomes Vout=Vin−Vgs. The voltage Vgs between the gate and the source of the amplifying transistor 401 is equal to the bias electric potential Vb, and hence the input/output relationship becomes Vout=Vin−Vb. However, as shown in FIG. 5, this equation is only valid when the biasing transistor 402 operates in the saturated region (corresponds to the case when Vin is large). In the case Vin is small and the biasing transistor 402 operates in a linear region, the equation Vout=Vin−Vb cannot become valid as shown in FIG. 6. When the biasing transistor operates in the linear region, the input/output relationship becomes Vout Vin−Vb′. The Vb′ here is the voltage between the gate and the source of the amplifying transistor 401 at that point. If the electric current flowing in the biasing transistor 402 is Ib′ when the biasing transistor 402 is operating in the linear region, then Ib′<Ib. Consequently, the relationship between Vb and Vb′ becomes Vb′<Vb. In other words, when Vin and Ib′ becomes small, then Vb′ also becomes small. As a result, the input/output relationship (relationship between Vin and Vout) becomes non-linear as shown in FIG. 7.
The following fact can be discerned from the above explanation.
First, to increase an amplitude value of the output electric potential Vout in the source follower circuit, it is appropriate to make the bias electric potential Vb small. Since Vout=Vin−Vb, when Vb is small, the Vout can be increased. However, it is necessary that the biasing transistor 402 be in conductive. Therefore, the value of the bias electric potential Vb must be made larger than that of a threshold voltage of the biasing transistor 402.
To the contrary, in the case the bias electric potential Vb is large, the biasing transistor 402 can readily operate in the linear region when the input electric potential Vin becomes small. As a result, the input/output relationship of the source follower circuit is likely to become non-linear. It is appropriate, in this respect, to make the bias electric potential Vb small.
The operation of the source follower circuit under a fixed normal state has been explained so far. Next, the operation of the source follower circuit under a transient state will be explained. As a circuit structure thereof, the circuit shown in FIG. 4 will be used with the addition of a load. In other words, the circuit structure here is a structure in which a load capacitance 805 is connected between output terminals, that is, a source terminal of an amplifying transistor 801 and a load capacitance power source line 806 as shown in FIG. 8. Therefore, the electric potential of the load capacitance 805 is the same as the output electric potential Vout of the source follower circuit.
First, a case where the output electric potential Vout is small in the initial state, that is, when Vout<Vin−Vb. FIG. 8A is a diagram showing a circuit configuration, and FIG. 8B is a diagram showing a timing chart. In that case, a value of a voltage Vgs between a gate and a source of an amplifying transistor 801 is larger than a value of a voltage Vgs between a gate and a source of a biasing transistor 802. Therefore, a large electric current flows in the amplifying transistor 801, and as a result, a load capacitance 805 is rapidly charged and the output electric potential Vout becomes large, whereby the voltage Vgs between the gate and the source of the amplifying transistor 801 becomes smaller. When the voltage Vgs between the gate and the source of the amplifying transistor 801 finally becomes equivalent to the bias electric potential Vb, the transient state is turned into a fixed normal state. The output electric potential Vout at that point is Vout=Vin−Vgs=Vin−Vb. Thus, as in the case where Vout<Vin−Vb, initially the voltage Vgs between the gate and the source of the amplifying transistor 801 is large under the transient state. Therefore, a large electric current, passing through the amplifying transistor 801, flows to the load capacitance 805. The writing-in time of a signal to the load capacitance 805 can thus be performed in a short time.
On the other hand, a case is discussed where the output electric potential Vout is large in the initial state, that is, when Vout>Vin−Vb. FIG. 9A is a diagram showing a circuit configuration thereof, and FIG. 9B is a diagram showing a timing chart thereof. In that case, because a value of a voltage Vgs between a gate and a source of an amplifying transistor 901 is small, the amplifying transistor 901 is in a non-conductive state. Then, the electric charges that have accumulated in a load capacitance 905 flow through a biasing transistor 902 to thereby be discharged. At that point, a voltage between a gate and a source of the biasing transistor 902 is the bias electric potential Vb, and therefore the electric current flowing in the biasing transistor 902 becomes Ib. As the output electric potential Vout gradually becomes smaller, the voltage Vgs between the gate and the source of the amplifying transistor 901 becomes larger. When the voltage Vgs between the gate and the source of the amplifying transistor 901 finally becomes equivalent to the bias electric potential Vb, the transient state is turned into the fixed normal state. Under the fixed normal state, the value of Vout is a fixed value, and hence an electric current will not flow in the load capacitance 905. The electric current Ib will continuously flow in the 2 transistors of the source follower circuit.
Thus, from the above explanation, it can be understood that when Vout>Vin−Vb, the electric discharging time of the load capacitance 905, that is, the signal writing-in time is determined by the electric current Ib flowing through the biasing transistor 902. The amount of the electric current Ib is determined by the size of the bias electric potential Vb. Therefore, to increase the electric current in order to shorten the signal writing-in time to the load capacitance 905, it is necessary to increase the bias electric potential Vb.
Next, a timing chart of a signal in a pixel 309 is shown in FIG. 10. First, the resetting transistor 307 is turned into a conductive state by controlling the reset signal line 305, whereby the electric potential of the N channel side terminal of the photo diode 304 is charged until the electric potential Vd of the power source line 309. In other words, the pixel is reset. Subsequently, the resetting transistor 307 is turned into a non-conductive state by controlling the reset signal line 305. Thereafter, when light is irradiated to the photo diode 304, an electric charge according to the light density is generated. Therefore, the electric charge that is charged due to the resetting operation is gradually being discharged. In short, the electric potential of the N channel side terminal of the photo diode 304 decreases. In the case a dark light is irradiated to the photo diode 304, the amount of electric discharge is small, and therefore the electric potential of the N channel side terminal of the photo diode 304 does not decrease much. Then, at a certain point, the switching transistor 301 is turned into a conductive state to thereby read-out the electric potential of the N channel side terminal of the photo diode 304 as a signal. This signal is proportional to the density of light. Then, the resetting transistor 307 is turned into the conductive state again to thereby reset the photo diode 304, and similar operations are repeated.
A transistor in the pixel 309 will be explained next. Regarding the polarity of the transistor thereof, all are N channel types most of the time. In rare cases, a P channel type may be used for the resetting transistor (JIEC Seminar: Development Prospects of the CMOS Camera: Feb. 20, 1998, refer to FIG. 11). Further, with regard to a method of lining up (arranging) the amplifying transistor and a selecting transistor, N channel types are used for both transistors and as shown in FIG. 3, often the structure is one in which the power source line 309 and the amplifying transistor 306 are connected, the amplifying transistor 306 and the switching transistor 301 are connected, and the switching transistor 301 and the signal output line 303 are connected. In rare cases N channel types are used for both transistors and the structure thereof is one in which the power source line 309 and the switching transistor 301 are connected, the switching transistor 301 and the amplifying transistor 306 are connected, and the amplifying transistor 306 and the signal output line 306 are connected (ISSCC97: p. 180, A ¼ Inch 330K Square Pixel Progressive Scan CMOS Active Pixel Image Sensor).
Next, a sensor portion for performing photoelectric conversion or the like will be explained. A PN type of photo diode is usually used to convert light into electricity. However, there are other types including a PIN type diode, an avalanche diode, an NPN incorporated diode, a Schottky diode, etc. There are also others such as a photo diode for X-rays and a sensor for infrared rays. These are described in “The Basics of Solid Imaging Elements: DENSHINO MENO SHIKUMI” written by Takao Ando and Hirohito Kobuchi: Nippon Riko Shuppan Kai.
Products suitable as sensors will be explained next. Other than the digital still camera and scanner, a sensor may also be used in an X-ray camera. In that case, there is a case where the photo diode for directly converting an X-ray into an electric signal is used or a case where an X-ray is converted into light by using a fluorescent material or a scintillator and then the light is read. The case of converting an X-ray into light by using a scintillator and thereafter reading the light is described in “Euro Display 99: p. 203: X-ray Detectors base on Amorphous Silicon Active Matrix”. In the “IEDM 98: p. 21: Amorphous Silicon TFT X-ray Image Sensors”, a case of reading light by using an amorphous silicon is reported, and a case of reading light by using a photo conductor is reported in the “AM-LCD99: p. 45: Real-time Imaging Flat Panel X-ray Detector”.
First, consideration is made on the item required in a source follower circuit 405. The most necessary item is to obtain a value as large as possible as an amplitude of the output electric potential Vout, that is, a value that is roughly equivalent to an amplitude of the input electric potential Vin. If the amplitude of the output electric potential Vout is large, signals having a large number of gradations can be obtained. As a result, the quality of the image read from an image sensor is enhanced. In addition, it is necessary that the input/output relationship is linear. In other words, it is crucial that the relationship of the input electric potential Vin and the output electric potential Vout in the source follower circuit operate linearly in a wide range. That is, the relationship of Vout=Vin−Vb is maintained even if the input electric potential Vin is small. In short, it is important that the biasing transistor 402 operate in the saturated region. Other items that are necessary include a short signal writing-in time of the output electric potential Vout to the load capacitance. If the signal writing-in time is long, the operation thereof will become slow.
Then, consideration is now made regarding a method to satisfy the above-mentioned items required in the source follower circuit.
First, because Vout=Vin−Vb, it is appropriate to make the bias electric potential Vb small in order to increase the amplitude of the output electric potential Vout. Similarly, the bias electric potential Vb may be made small in order to widen the operating region of a linear input/output relationship. The reason for this resides in that when the bias electric potential Vb is small, the biasing transistor 402 can easily operate in the saturated region even if the output electric potential has become small. However, when the bias electric potential Vb is small, the writing-in time of the output signal becomes long.
In other words, the amplitude of the output electric potential and the signal writing-in time have a trade-off relationship. It is impossible to shorten the writing-in time of the output electric potential while increasing the amplitude value of the output electric potential. In addition, it is also impossible to widen the operating region in which the input/output relationship is linear while increasing the amplitude value of the output electric potential.